Create a 21-slide Master's Thesis defense presentation for the evaluation panel at the Autonomous University of Barcelona (UAB). The audience is composed of academics and experts in remote sensing, geography, and agronomy. They are familiar with land-cover mapping but are skeptical about operational crop monitoring in tropical, cloud-prone, fragmented landscapes using optical imagery. They expect rigorous methodology, clear evidence of field validation, and honest discussion of limitations. After the presentation, the panel should: Understand that Sentinel-2 time-series, when strategically sampled, can discriminate four key crops (maize, sugarcane, musaceae, watermelon) in Honduras. Recognize the novelty of the hybrid variable-selection strategy (Gini importance + phenological coherence) that reduces predictors from 90 to 25 while achieving 96.11% overall accuracy. Be convinced that the method is replicable, open-access, and ready for adoption by national institutions (SAG, ICF). Ask informed questions about the maize-watermelon confusion (JM=1.92) and the critical November temporal window that resolves spectral convergence. Preferred number of slides: ~21 (one per main component, matching the structure of the uploaded content). Video duration: ~15-20 minutes (defense style). Language: English (US).
Create a 21-slide Master's Thesis defense presentation for the evaluation panel at the Autonomous University of Barcelona (UAB). The audience is composed of academics and experts in remote sensing, geography, and agronomy. They are familiar with land-cover mapping but are skeptical about operational crop monitoring in tropical, cloud-prone, fragmented landscapes using optical imagery. They expect rigorous methodology, clear evidence of field validation, and honest discussion of limitations.
After the presentation, the panel should:
Understand that Sentinel-2 time-series, when strategically sampled, can discriminate four key crops (maize, sugarcane, musaceae, watermelon) in Honduras.
Recognize the novelty of the hybrid variable-selection strategy (Gini importance + phenological coherence) that reduces predictors from 90 to 25 while achieving 96.11% overall accuracy.
Be convinced that the method is replicable, open-access, and ready for adoption by national institutions (SAG, ICF).
Ask informed questions about the maize-watermelon confusion (JM=1.92) and the critical November temporal window that resolves spectral convergence.
Preferred number of slides: ~21 (one per main component, matching the structure of the uploaded content).
Video duration: ~15-20 minutes (defense style).
Language: English (US).
Created using ChatSlide
This project focuses on multitemporal crop analysis in Honduras, led by Gabriel Banegas-Rodríguez at UAB, aiming to demonstrate how phenology can enhance crop discrimination. The research investigates whether phenology can effectively separate tropical crops by testing various time windows and selection strategies across four key food crop classes. Additionally, it addresses the monitoring gap in Honduras, highlighting the limitations of national maps that aggregate short-cycle crops and the...